« Infrastructure Overbuilds: Past and Present | Main | Hume's Sailing Ships correcting Samuelson's Icebergs »


Feed You can follow this conversation by subscribing to the comment feed for this post.

Well. Yes and no. The textbook downward-sloping AD curve in P-Y space is just wrong, I think. It depends on the money stock being a slow variable and the price level being a fast variable and I don't think there's any question about modern economies that's best thought of in those terms.

The textbook AS curve is defensible, we can think of it as representing any of various stories about how eliciting greater material output in a given institutional framework requires/provokes a general rise in the price level. The conventional wage-bargaining-plus-markup-pricing story is one acceptable way to get this.

In place of the AD curve we could put an upward sloping 45-degree line, representing the fact that higher money-spending must result in some mix of higher prices and higher real output. We definitely do not want a downward sloping "AD" curve. The only function of this is to encourage misleading analogies from micro.

But we really have to ask, what questions do we want students thinking about? You could take the model above and extend it with an IS curve and close it with the identity that real interest rates = nominal less inflation. That would focus the attention on Wicksell's cumulative process - without a central bank actively adjusting rates, and deviation in demand creates a positive feedback carrying you away from Y* (however defined). Personally I don't think that's a bad lesson for students to take away from a macro class. If you want a different lesson, you'll want a different model.

I don't see what's wrong with Y = C + I + G + NX.

JW: the shape of the AD curve depends very much on what the central bank is targeting. (I had an old post on this.) If it targets M (like in Friedman's k% rule) it should be downward-sloping. An NGDP target should make it a rectangular hyperbola. But with the wrong target it could easily slope the wrong way. Like as you say, a nominal interest rate target makes the AD curve vertical in {P,Y} space, and slope the wrong way in {inflation,Y} space. Which matters, for the reasons you say.

"The only function of this is to encourage misleading analogies from micro."

That worries me. But I think that danger is there whether or not we do AD/AS, and we need to confront it head on: "why are the AD/AS curves different from the D/S curves for apples?" "Will they slope the right ways so price flexibility gets us to full employment?"

There's nothing wrong with Y=C+I+G+NX, but it's not a theory of anything, just a rather arbitrary way of dividing up categories. And why the focus on newly-produced goods, when *all* trade is (presumably) beneficial to buyer and seller?

Put it this way: if each price is flexible it may give us D=S in each market; but if all prices are flexible will that give us D=S in all markets? Drawing the AD/AS curves, and asking why they slope the way they do, forces us to face that question.

In the red/green world I can increase my flow out today to reduce my flow out tomorrow. Or vice versa. How does that affect AD/AS? Does it matter?

Miami: I think that's true in a green only world as well. Think it means AD shifts right today and left tomorrow.

Sort of but, not exactly? In the green only world I can increase my flow out today to increase my flow in tomorrow. Different flows. Whether that makes a difference idk.

the shape of the AD curve depends very much on what the central bank is targeting.

This raises the question of whether the central bank should be inside the model or outside the model. Do we want students to ask "given what we know about the economy, how would we like the central bank to behave?" Or do we want them to ask "Given what we know about the economy and about the central bank's actions, which we have no say in, how would we like our elects government to behave?"

Nick Rowe,

Thanks for sharing this post! I think teaching aggregate demand is the hardest part of teaching introductory macro (especially for a micro guy like myself).

Have you ever wondered if the task would be easier if we called the AD curve a "money market equilibrium" curve, instead? Because doesn't the AD curve really just show the combinations of P and Y that equilibrate the money market? That way there is no confusion with micro demand curves?

That is sort of the approach I was thinking of taking the next time I taught macro. Here are my notes if you want to take a look:

JW: Hmm. Dunno. My preferred is a sort of mix. Run the model with different assumptions about how the central bank behaves, then ask which which of those we like.

DW: Thanks! Teaching intro macro is hard, even for old macro guys like me. I really feel for micro people like you, and applaud you, because it's the most important teaching job in economics (hell, in the university!).

I see your point about calling it a "money market equilibrium curve". And it *is* a *semi*-equilibrium curve, because it shows points at which Yd(M,P,Y)=Y. But not *full* equilibrium (which is where Yd=Y=Ys). But I dislike the phrase "money market", because *all* markets are "money markets" in a monetary exchange economy. (The "apple market" is a market where apples are traded for money, etc.).

But you are exactly right in the sense that the AD curve shows combinations of P and Y at which people do not want to change their stocks of money by increasing or decreasing the flow of money *out* of their pockets.


Thanks for the response! I think you make a good point that AD is only a semi-equilibrium curve (since full equilibrium requires AS too).

I guess I am just trying to think of ways to emphasize the fact that micro demand curves illustrate a behavioral relationship between the relative price for a good and the quantity demanded of that good. Namely, if you lower relative prices for a good, people will want to buy more of it.

By contrast, AD curve illustrates relationship between price level and total spending on output. But this isn't really a behavioral relationship in the sense that lower price levels lead people to "want" to spend more on output. Instead, it's just that when the price level is lower, the output level that is associated with a Md=Ms is higher.

Would you agree that's the right distinction to make between micro and macro demand curves or am I steering my students wrong? Sorry to bug you with undergrad questions!

DW: This is exactly the right place to "bug" me with those "undergrad" questions.

You are steering your students right. But keep in mind the flow out/flow in distinction when you talk about "Md=Ms" (And it would be better to talk about "Md=M" rather than "Md=Ms", because that opens up another can of worms.) Because if we are on the AD curve but to the left of the AS curve, people want a higher flow in than the actual flow in (they want to sell more goods/labour than they can actually sell).

By the way, as a microeconomist, you might be interested in my minimalist model which explains macro in terms of a 3D Edgeworth box. Because barter is ruled out, you can't trade diagonally.

M is not demand, it is Money. How do you teach about the creation of money itself?

Of course there is no demand unless there is money, so M equaling AD has some explanatory meaning. But it seems to me you first need to teach how money is created. Skipping this step in the modeling and teaching is missing a very important part to an understanding of how modern economies work.

One oart of the class might take a view as if they were China, another part could think of themselves as Being in the US, another part could think as if they were the european union.

According to today's press this discussion about money and credit creation has become important at the Davos conference. Akways good to have some current events to go along with the financial crisis itself to help in teaching.


This minimalist model is very interesting! I feel like I learned something anyways. Will def link this to my students the next time I teach macro. Thanks for posting this as well!


Pr Rowe,

Woulsd you mind to answer two naive questions from a non economist ?

#1 In your minimalist model in what "Recessions are not about output and employment and saving and investment and borrowing and lending and interest rates and time and uncertainty. The only essential things are a decline in monetary exchange caused by an excess demand for the medium of exchange. Everything else is just embroidery." is different from the Keynesian liquidity trap concept ?

#2 4 stocks and two flows look like a graph. Graph don't seem to be used by macroecomists. Could this kind of tool be useful for transcations study ? (An economic transaction is not that different that a transaction made between two computers in a network). I've asked this question some time ago on Noah Smith blog but did not got any clear answer and I guess that people having to deal with graph probably ask themselves the same question.


Ludovic Coval

A great, simple model. Thanks Nick
Here is a simple explanation. You have a large set of baskets, big and small. And you also have a savings and loan machine.
When the large baskets empty into the small baskets, large deposits occur in the machine, counter balanced by small loans. And visa versa. Each basket maintains an account with the S&L, and there are no fees.

When the basket flow changes, the machine notices the change, and re-balances the deposits and loans by forced, asynchronous partial amortization. If the basket have multiple sizes, hey will be a bucket brigade, a supply chain. The S&L will have a yield curve to match with the number of points on the yield curve equivalent to the number of basket sizes needed to prevent queuing up. .

The equilibrium condition is when all baskets are generally about half full, say. Then when the basket moods strikes, the baskets can skip the long queues and jump on the short ones, be opportunistic. They have surplus. Hence the economy flows without pileups.

Let me add one more remark, please.
The model is very interesting because the driver of a dairy truck can count gallons per day, a fairly remarkable feat, made possible because he has dollars per delivery. All of this algebra, its mere existence, is about half of your aggregate macro economics theory.

The connection between the S%L couple basket brigades is simple. The S&L, acting as above, forces a container algebra, containers ratio up, they have prime numbers. Example, the grocery plastic hand carry case, we see them all the time. Its prime, an entire industry stacks those up into pallets, panel trucks, flatbeds, train cars and so one. All of those container sizes compose and decompose with integer arithmetic of small order.

But, by keepng them half full, keeping the queues baklanced, the containers can be extrapolated into linear compressible flow, like silly putty. The smooth arithmetic works because the S&Ls keep track of bit error each time we swipe the credit card in an exchange.

JF: normally in teaching intro macro there's a separate chapter on why people use money (rather than barter) and on how central banks and the banking system create money. Then you do AD/AS.

DW: I am really pleased that someone like you finds this stuff useful! I wanted to tie macro in with micro, at a very basic level. Micro teaches us that there are gains from trade; the pure endowment economy 2D Edgeworth Box is the simplest example. Macro adds monetary exchange, and shows that an excess demand for money causes a fall in trade (a "recession"), which makes people worse off. Once students have grasped that basic idea, you can add all the bells and whistles later.

#1. The Keynesian Liquidity Trap is a very special case of the model, where if M increases it doesn't cure the recession. I think (haven't checked) that if I changed my minimalist model so that preferences became U=log(A)+log(B)+M (so I replaced "log(M)" with just "M") it would generate something like a liquidity trap.

#2. Macroeconomists do use graphs. It shouldn't be too difficult to portray my minimalist model in a 2D graph, though a 3D graph would be better, because there are 3 goods. It's a 3D Edgeworth Box.

Thanks Matt! But I think adding a savings and loan machine complicates it.

Pr Rowe,

Thanks for your answers.

I may be wrong but I suspect that we don't use the word graph with same meaning that is function graph vs Graph Theory. https://en.wikipedia.org/wiki/Graph_theory (Sorry I don't know HTML tags for URLs).

Ludovic: Yep. I misunderstood you.


"Normally in teaching intro macro there's a separate chapter on why people use money (rather than barter) and on how central banks and the banking system create money. Then you do AD/AS"

What level of flexibility is afforded to the central bank in introductory macro? Presumably in intro macro, the central bank is precluded from purchasing real goods and that money creation is limited to a credit channel?

If central banks are limited to operating in credit channels, and with private banking, shouldn't there also be an AD/AS set of curves for credit based money where the price of credit is the interest rate? I think this is the heart of JF's question - don't you need AD/AS curves for all goods (including money) for a complete model?

With a simple model where only the central bank exists, lending money to any economic actor, the price of credit can be lowered to any point necessary to reach the desired monetary level. With private banking, it seems there is a some minimum positive interest rate that private banks will demand to meet it's own cash flow considerations (paying employees, shareholders, etc.).

Nick, this model is actually included in Mankiw's macro textbox. In one of the first chapters he first derives the AD curve using the monetarist approach with the equation of exchange. Then in a later chapter he derives it using IS/LM.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad